Priority Effects Are Interactively Regulated by Top-Down and Bottom-Up Forces: Evidence from Wood Decomposer Communities

Home , Fungivore

Ecology Letters, (2017) doi: 10.1111/ele.12803 LETTER Priority effects are interactively regulated by top-down and bottom-up forces: evidence from wood decomposer communities

Abstract Devin R. Leopold,1* J. Paula Both top-down (grazing) and bottom-up (resource availability) forces can determine the strength Wilkie,2 Ian A. Dickie,3,4 Robert B. of priority effects, or the effects of species arrival history on the structure and function of ecologi- Allen,5 Peter K. Buchanan6 and cal communities, but their combined inﬂuences remain unresolved. To test for such inﬂuences, we Tadashi Fukami1* assembled experimental communities of wood-decomposing fungi using a factorial manipulation of fungivore (Folsomia candida) presence, nitrogen availability, and fungal assembly history. We found interactive effects of all three factors on fungal species composition and wood decomposition 1 year after the fungi were introduced. The strength of priority effects on community structure was affected primarily by nitrogen availability, whereas the strength of priority effects on decomposition rate was interactively regulated by nitrogen and fungivores. These results demon- strate that top-down and bottom-up forces jointly determine how strongly assembly history affects community structure and function.

Keywords Assembly history, fungivore grazing, historical contingency, priority effects, resource availability, saprotrophic fungi, wood decomposition.

Ecology Letters (2017)

species (Fukami et al. 2016). Greater resource availability gen- INTRODUCTION erally strengthens priority effects by increasing the growth of The order and timing of species arrival during community early-arriving species (Chase 2003; Fukami 2015; but see assembly can affect species composition (Palmgren 1926; Fukami et al. 2010; Dickie et al. 2012), as shown in plants Sutherland 1974; Drake 1991) and ecosystem functioning (Kardol et al. 2013), animals (Chase 2010), and microbes (Korner€ et al. 2008; Tan et al. 2012), the phenomenon known (Vannette & Fukami 2014). as priority effects. Because species arrival order is often highly Although both top-down and bottom-up forces determine stochastic and unknown, priority effects introduce historical assembly history effects, their potential interactive effects contingency in community assembly, which complicates the remain unclear. In a pioneering study of grassland plant com- search for general patterns in communities (Diamond 1975; munities, Ejrnæs et al. (2006) found that interactions between Lawton 1999). However, the strength of priority effects is species introduction order, nutrient availability, and simulated known to vary with ecological context (e.g. community size, herbivory (manual defoliation) determined species richness Orrock & Fletcher 2005; phylogenetic diversity, Tan et al. and composition. In that study, however, introduction order 2012; disturbance, Tucker & Fukami 2014). Progress towards a was limited to two treatments that were confounded with better understanding of community composition and ecosystem plant type (specialist vs. generalist) and the simulated her- function can therefore be made by identifying conditions that bivory manipulation was applied equally to all species. So far, make assembly history important (Chase 2003; Fukami 2015). no studies, to our knowledge, have addressed the possibility Factors affecting the strength of priority effects include both that the relationship between assembly history and emergent top-down forces (e.g. grazing) and bottom-up forces (e.g. functional properties of communities depend on interactions resource availability). Grazing by consumers can weaken pri- between bottom-up and top-down forces. ority effects by reducing population growth of early-arriving In this paper, we experimentally test for the joint effects of species (Morin 1984) or by altering the competitive hierarchy top-down and bottom-up forces on priority effects. To this among species (Louette & De Meester 2007; Chase et al. end, we assembled communities of wood-decomposing fungi 2009). It is also possible for consumers to strengthen priority in laboratory microcosms, manipulating species arrival order, effects if they limit the effective size of a local community initial nitrogen availability, and the presence of fungivorous (Chase et al. 2009) or equalise ﬁtness differences among spe- springtails (Collembola). In wood-decomposing fungi, the his- cies by preferentially consuming a competitively dominant tory of species arrival can vary among communities (e.g.

1Department of Biology, Stanford University, Stanford, CA 94305, USA 5Independent Researcher, Lincoln 7608, New Zealand 2Independent Researcher, Auckland 1025, New Zealand 6Landcare Research, Auckland 1142, New Zealand 3BioProtection Research Centre, Lincoln University, Lincoln 7647, New Zealand *Correspondence: [email protected], [email protected] 4University of Canterbury, School of Biological Sciences, Christchurch 8140, New Zealand

among fallen logs) owing to variation in sporulation timing, and 400 mL water or water with NH4NO3 (see below). Wood mycelial growth, the phenology of insect vectors, and the tim- disks (c. 8 cm diameter by 1 cm thick) were prepared from ing of tree death or log fall (Boddy et al. 2017). This varia- freshly felled N. solandri trees and dried to constant weight to tion, coupled with strong local interactions (Boddy 2000), can determine their initial dry mass. The wood disks were then make priority effects a major driver of community assembly soaked in water for 48 h and placed on top of the soil in the (Fukami et al. 2010). In addition, both increased nitrogen microcosms. Before inoculation with fungi, the assembled availability and fungivore grazing are known to influence the microcosms were capped with loosely fitting lids and auto- composition and function of fungal communities (Wardle & claved twice at 121 °C with a 24 h interval. Experimental Yeates 1993; Crowther et al. 2012; Morrison et al. 2016), yet treatments were randomly assigned to jars, which were then their potential to interactively regulate priority effects is unex- randomly arranged on laboratory shelves and maintained in plored. the dark at c.20°C and 60% humidity throughout the course As resource availability may affect fungal communities more of the experiment. strongly than grazing (Zimmerman et al. 1995; Wardle 2002; Moore et al. 2003), we predicted that nitrogen addition would Fungal species influence the strength of priority effects more than fungivore presence. However, because Collembola may also be resource- The fungi used in this study were isolated from the same site limited initially (Booth & Anderson 1979; Larsen et al. 2011), as soils (above). The 10 species used in this study (Fig. S1) we also predicted that the influence of fungivores on the were largely identical to those used by Fukami et al. (2010) strength of priority effects will be greater with, rather than and Dickie et al. (2012), which were selected from the initial without, nitrogen addition due to more rapid population collection of 96 species to maximise ease of handling, phyloge- growth in response to increased fungal resource availability. netic diversity, and adequate molecular discrimination using This study builds on earlier experimental work linking priority nrDNA internal transcribed spacer (ITS) length heterogeneity. effects to functional properties of wood-decomposing fungal In preliminary experiments, we found that one species used in communities (i.e. decomposition rate and carbon and nutrient the previous studies, Phlebia nothofagi, caused rapid and com- dynamics; Fukami et al. 2010; Dickie et al. 2012). The novelty plete mortality of Collembola in vitro. Because P. nothofagi of the present work is that, unlike any previous research we are was a final dominant species in previous assembly experiments aware of, manipulations of both top-down and bottom-up fac- (Fukami et al. 2010), we excluded it from the current study to tors were combined with manipulations of species arrival order. ensure that we could effectively manipulate fungivore presence. Instead of this species, we used Helicogloea alba (New Zealand Fungarium, PDD 91620; ICMP 17042; GenBank MATERIALS AND METHODS accession GQ411522.1) in the current study. Selecting species that are amenable to manipulation in pure Overview and experimental design culture may have reduced functional diversity relative to the Fungal communities, consisting of 10 species, were assembled broader species pool in the field. It is also possible, however, on sterile wood disks in individual microcosms at one of two that efforts to select a phylogenetically diverse group of spe- initial nitrogen levels and in the presence or absence of cies increased the functional diversity of the microcosm com- Collembola. Assembly history was manipulated by introduc- munities relative to the diversity of initial colonists under ing one of four randomly selected initial fungal species natural conditions. The ten species represent a range of wood- 4 weeks prior to introducing the remaining nine species. A decomposition strategies, including white-rot, brown-rot, and control treatment, where no fungi were introduced was also soft-rot fungi (Worrall et al. 1997; see Fig. S1). included. The experimental design was fully factorial, with 5 replicates of each treatment destructively harvested at 6 and Assembly history 12 months, resulting in 200 samples (i.e. 2 nitrogen levels 9 2 fungivore treatments 9 5 assembly histories (including con- Assembly history treatments were established by first inoculat- trol) 9 5 replicates 9 2 harvests, as detailed below). To aid ing sterilised wood disks with one of four randomly selected interpretation of the fungal community-level responses to the species, either Ascocoryne sarcoides, Bisporella citrina, Dal- experimental treatments, a second set of microcosms were dinia novae-zelandiae, or Trametes versicolor. The feasible inoculated with only one fungal species, resulting in 400 addi- number of assembly history treatments was limited due to tional, single-species replicates (i.e., 10 species 9 2 nitrogen practical constraints on the number of samples needed for the levels 9 2 fungivore treatments 9 5 replicates 9 2 harvests). factorial design. Inoculation consisted of aseptically transfer- ring a 5-mm plug from a colonised agar plate to a predetermined location on the upper surface of each wood disk. The Microcosms remaining nine species were aseptically transferred to the Microcosms were prepared as in Fukami et al. (2010). Briefly, wood disk 4 weeks later. The inoculation positions for each mineral soil was collected from a monospecific stand of old-- species, relative to the other species, was invariant across all growth Nothofagus solandri (Allen et al. 2000; Clinton et al. treatments to avoid potential confounding effects of spatial 2002) at Craigieburn Forest Park, South Island, New Zealand arrangement (Fig. S1). We used the 4-week interval to allow (43°8.556 S, 171°42.825 E, 1000 m elevation) and added to the initial species to become established without completely 2 L glass jars. Each jar received 800 g of dry, well-mixed soil colonising the disk. It is uncertain how common this interval

© 2017 John Wiley & Sons Ltd/CNRS Letter Grazing, resources, and priority effects 3 may be under natural conditions. However, initial colonisa- wood-splitting device was used to avoid contamination of the tion by a single species, followed by simultaneous colonisation interior wood, as detailed in Fukami et al. (2010) and Dickie by numerous other species after some time interval, is plausi- et al. (2012). Approximately 5.5 mg of sawdust was removed ble. For example, a dead branch that is colonised by one fun- from the interior wood at nine predetermined locations gus while still attached to a tree may then fall to the ground (Fig. S1) using a flame-sterilised 1.5-mm drill bit, and col- and simultaneously come in contact with multiple additional lected in individual sterile 0.2 mL tubes. species (Boddy et al. 2017). To characterise fungal species composition, each of the nine sawdust samples from each disk were analysed separately using a modified length-heterogeneity polymerase chain reaction Nitrogen availability (LH-PCR) assay. First, total fungal DNA was extracted and We chose to manipulate nitrogen availability as a bottom-up PCR amplified using the Sigma REDExtract-N-amp Plant factor because fungal decomposition of highly lignified sub- PCR kit and the labelled primers ITS1F-6FAM and ITS4-VIC strates with a high carbon-to-nitrogen ratio, such as wood, is (Dickie et al. 2009). Fungi present in each PCR product were often nitrogen-limited (Allison et al. 2009; Boberg et al. then identified by analyzing ITS fragment lengths with 50-cm 2014), although fungal species may vary in their response to capillary electrophoresis in a 3130xl Genetic Analyzer (Applied nitrogen. The mineral soil collected from the field site and Biosystems, Foster City, CA, USA). Peak profiles were com- used in the microcosms was nutrient-poor and primary pro- pared against a database of the known ITS sequence lengths ductivity at the site was nitrogen-limited (Clinton et al. 2002; and converted to species presence data for each sawdust sam- Davis et al. 2004; Smaill et al. 2011). Microcosms assigned to ple using TRAMPR (FitzJohn & Dickie 2007). the low-nitrogen treatment received no supplemental nitrogen Functional outcomes of fungal community assembly were and reflected the natural nutrient availability in the wood characterised by measuring the wood mass lost through decom- disks and mineral soil. The high-nitrogen treatment received position and the final carbon and nitrogen concentrations in 1gNH4NO3 in the 400 mL water added to each microcosm the wood disks. Following sampling for molecular analysis, the before autoclaving. The resulting range of soil nitrogen avail- wood disks were dried at 40 °C to a constant weight and the ability in the experimental treatments reflected the range of mass was recorded. The change in dry mass for each disk was values naturally present in soils where the fungi were col- used as a proxy for decomposition, but also includes fungal tis- lected, including the relatively nutrient rich organic soil layers sue. In a laboratory experiment, Jones & Worrall (1995) found (Clinton et al. 2002), and coincides with the high- and low- that 3–30% of the dry biomass remaining after 12 weeks was nitrogen treatments used by Fukami et al. (2010). likely fungal. Although this study is not directly comparable to ours, it indicates that fungal biomass might be relatively minor compared to wood mass. The mass of sawdust removed for Fungivore presence molecular analysis was not accounted for in the final dry We manipulated the presence of fungivorous mesofauna in weights, but was negligible relative to mass lost to decomposi- the experimental microcosms by adding a species of Collem- tion and was consistent across treatments. Carbon and nitro- bola, Folsomia candida. This species was chosen because it is gen were analysed after determining the final dry mass of each primarily mycophagous and could be easily reared under lab- disk by collecting a composite sawdust sample from new loca- oratory conditions, which helped to avoid unintended intro- tions along each of the eight split edges using a 2-mm drill bit. duction of contaminant fungi to the microcosms (Fountain & In addition, immediately before the harvest, we visually Hopkin 2005). In addition, F. candida has been previously inspected the microcosms and recorded the abundance of used in studies of the ecological impacts of grazing by soil F. candida ordinally. We classified each microcosm into one fungivores (e.g. Lussenhop 1996; Cragg & Bardgett 2001). To of four categories (i.e. 0 = no F. candida individual in the ensure that the initial fungal species had an opportunity to microcosm; 1 = 1–10 individuals; 2 = 10–100 individuals; and establish on the wood disk and that there would be sufficient 3 = more than 100 individuals). The F. candida population resources for the F. candida population, c. 30 individuals of persisted until the end of the experiment in most microcosms the fungivores were aseptically transferred into each of half of (Figs S2 and S3). the microcosms belonging to each assembly and nutrient treatment 4 weeks after the final fungal inoculation. It should Statistical analysis be noted, however, that fungivores can be present throughout fungal community assembly under natural conditions. To determine whether the effects of assembly history were regulated interactively by nitrogen availability and fungivore grazing, we tested for significant three-way interactions of Harvest and data collection these experimental treatments as predictors of fungal commu- Microcosms were destructively harvested at one of two time nity composition and function. To quantify fungal species points, 6 or 12 months after the introduction of the initial composition, we used, as a measure of species prevalence in fungal species, to determine if priority effects were persisting the wood, the numbers of subsamples from each wood disk in or attenuating over time. Specifically, each wood disk was which each species was observed to construct a species by removed from the jar within a laminar flow hood and split sample matrix of relative abundance. The relationships along eight radial lines at predetermined locations relative to between species composition and experimental treatments the initial inoculation points (Fig. S1). A custom, sterile were visualised using non-metric multidimensional scaling

(NMDS) and the significance of these relationships were Fungal community composition tested with a permutational multivariate analysis of variance (perMANOVA) using the adonis function in the R-package The composition of the fungal community after 12 months var- vegan (Oksanen et al. 2017). For both the NMDS and perMA- ied with assembly history, initial nitrogen availability, fungivore NOVA analyses, community dissimilarity was calculated using presence, and their interactions (perMANOVA, F15,64 = 9.32, the modified Gower dissimilarity index proposed by Anderson P < 0.001; Fig. 2a), including a significant effect of the three- et al. (2006). In addition, to examine how experimental treat- way interaction (F3,64 = 2.246, P = 0.019). The greatest amount ments affected individual species, the proportion of subsam- of variation in community composition was explained by ples per microcosm where a species was detected was assembly history (R2 = 0.36), followed by nitrogen availability modelled as a binomial response variable using generalised (R2 = 0.21) and fungivore presence (R2 = 0.11). Beta-dispersion linear models. The full model for each species was simplified among assembly history treatments (i.e. the strength of priority using reverse model selection and likelihood-ratio chi square effects; Fig. 2b) was reduced with nitrogen addition tests. (F1,73 = 16.59, P < 0.001), but was unaffected by fungivore To test whether the strength of priority effects on species presence (F1,73 = 0.153, P = 0.69) or the interaction between composition was influenced by our experimental treatments, nitrogen and fungivores (F1,73 = 0.038, P = 0.85). we compared the compositional dissimilarity among assembly Species varied in complex ways in their response to the history treatments (i.e. a measure of the variation in species experimental treatments, both in the mixed communities composition that can be attributed to species arrival order) (Fig. 3) and in the single-species microcosms (Fig. 1). For for each factorial combination of nitrogen availability and example, the proportion of wood disk subsamples where fungivore presence. To quantify compositional dissimilarity, T. versicolor was observed was strongly influenced by assem- we first calculated the beta-dispersion among samples with the bly history, averaging 86.1% when arriving ahead of other same nitrogen and fungivore treatment using the modified species and only 3.9% otherwise, but was not significantly Gower dissimilarity index and the function betadisper in the influenced by fungivore presence or nitrogen availability. The R-package vegan. We then extracted the distance of each sam- abundance of Sistotrema brinkmannii also responded to ple to its group centroid and tested whether they were corre- assembly history, but only under low-nitrogen conditions and lated with the nitrogen and fungivore treatments and their did not appear to be influenced strongly by fungivore grazing. interaction, including the assembly history treatment as a ran- For A. sarcoides, B. citrina, Calocera sinensis, and D. novae- dom effect, using the function lme in the R-package nlme zelandiae, the effect of assembly history on abundance varied (Pinheiro et al. 2016). with both nitrogen availability and fungivore presence. For The effect of the experimental treatments on wood decompo- example, B. citrina was ubiquitous in the wood disk when sition were tested using analyses of variance (ANOVAs), with the arriving first under low-nitrogen conditions, irrespective of proportion of dry wood mass lost, the percent nitrogen, and grazing. However, under high-nitrogen conditions, B. citrina the carbon-to-nitrogen ratio as response variables. The carbon- colonised only 40 and 4.4% of the wood disk in the absence to-nitrogen ratio was log-transformed to better meet the and presence of fungivores, respectively. When arriving after assumption of normality. The full models, including all experi- another species, the prevalence of B. citrina depended on the mental treatments and interactions, were simplified for each initial species identity, nitrogen availability and fungivores. response variable using F tests. In order to link the properties For example, the prevalence of B. citrina when preceded by of the wood substrate to fungal community composition, the T. versicolor, relative to its prevalence when arriving after mass loss, percent nitrogen, and carbon-to-nitrogen ratio for T. versicolor, was dependent on the nitrogen and fungivore each sample were correlated with the NMDS ordination by treatments. Under low-nitrogen conditions, the prevalence of vector fitting, using the envfit function in the R-package vegan. B. citrina was reduced when T. versicolor arrived first, irre- Analyses of species relative abundance and functional out- spective of fungivore presence. Under high-nitrogen condi- comes were repeated for the single-species microcosms using tions, initial colonisation by T. versicolor had no effect on the the same statistical methods, with the exception that assembly prevalence of B. citrina when fungivores were absent, but, history was not included as a factor. All statistical analyses facilitated growth when fungivores were present. Three spe- were conducted in R v3.3.1 (R Development Core Team cies, Armillaria sp., H. alba and Pleurotus purpureoolivaceus, 2016). were not detected at all, and one species, Artomyces ( Clavi- corona) candelabrum, was only detected twice at 12 months and once at 6 months. RESULTS Patterns of fungal community composition and wood decom- Wood decomposition position were qualitatively similar at both harvest time points and statistical support for treatment effects after 12 months The physical and chemical properties of the decomposing were as strong as, or stronger than, those observed at wood disk were correlated with assembly history, initial nitro- 6 months. We will focus here on the 12-month results and gen availability, fungivore presence and their interactions report the 6-month results as supplemental materials (Figs (Fig. 4). Significant three-way interactions were observed for S4–S7). When inoculated alone, all fungal species colonised wood mass loss (F3,64 = 8.99, P < 0.001), final nitrogen wood by 12 months, at least in the absence of fungivores and concentration (F3,64 = 11.36, P < 0.001), and the carbon-to- nitrogen addition (Figs 1, S5 and S7). nitrogen ratio (F3,64 = 4.96, P = 0.004). Fungal species

Ascocoryne Bisporella Calocera Daldinia Sistotrema Trametes (a) (b) (c) (d) (e) (f) 100 75 50 N * N ** N N * N N

substrate 25 F ** F F F * F F occupied (%) 0 NxF ** NxF NxF ** NxF NxF NxF (g) (h) (i) (j) (k) (l)

20 N N * N N N * N ***

loss (%) 10

wood mass F F F F F F ** 0 NxF * NxF NxF + NxF NxF NxF ** (m) (n) (o) (p) (q) (r) 0.15

0.10 N ** N *** N * N N *** N 0.05 F F F F F F *** nitrogen (%) 0.00 NxF NxF NxF NxF NxF NxF + (s) (t) (u) (v) (w) (x)

1000

C:N 500 N ** N *** N + N N *** N F F F F F F *** 0 NxF NxF NxF NxF NxF NxF * +N −F −N −F +N +F −N −F −N +F −N −F −N +F +N −F +N +F −N −F −N +F +N +F +N −F +N −F −N +F +N +F −N −F +N −F +N +F −N +F +N −F −N +F +N +F −N −F

−Nitrogen −Nitrogen +Nitrogen +Nitrogen −Fungivores +Fungivores −Fungivores +Fungivores

Figure 1 Effect of nitrogen addition (N) and fungivore presence (F) on wood volume occupied (a–f), % wood mass loss (g–l), % nitrogen (m–r), and carbon-to-nitrogen ratio (s–x) for single-species inoculations. Significant predictors are indicated for each response, where +P < 0.1, *P < 0.05, **P < 0.01, ***P < 0.001. See also Figs S5 and S7 for additional results for singe-species inoculations. composition in NMDS was correlated with these functional responsible for the effect of assembly history on nitrogen con- outcomes, based on envfit (mass lost, R2 = 0.44, P < 0.001; % N, centration and carbon-to-nitrogen ratio during decomposition R2 = 0.26, P < 0.001; C:N, R2 = 0.46, P < 0.001; Fig. S8). (Fig. 4e–l). However, in contrast to decomposition rate, the The effect of assembly history on decomposition rate nitrogen and carbon patterns in the single-species microcosms depended on nitrogen availability and fungivore presence (Fig. 1r and x) did not correspond clearly to the mixed species (Fig. 4a–d). Under low-nitrogen conditions, decomposition treatments (Fig. 4e–l). rate varied among assembly history treatments to a greater degree when fungivores were not present (Fig. 4a and b). DISCUSSION However, under high-nitrogen conditions decomposition rate was only minimally influenced by assembly history or fungi- Our results provide experimental evidence that nutrient avail- vores. This pattern can be largely explained by the response ability and fungivore presence interact with fungal inoculation of T. versicolor to the experimental treatments. In the single- history to determine species composition (Fig. 2) and wood species microcosms, decomposition by T. versicolor was decomposition (Fig. 4). Prior research has shown that fungi- about three times as quick as by any other species under low- vore grazing can alter competitive interactions among fungi nitrogen conditions without fungivores, twice as quick with by suppressing or stimulating hyphal growth, depending on the addition of fungivores, but only slightly quicker than the the species involved (Crowther et al. 2011a). Our results sug- across-species average under high-nitrogen conditions, with or gest that these effects of fungivores may depend on resource without fungivores (Fig. 1g–l). Because the prevalence of availability. For species composition, the joint effects of nitro- T. versicolor was strongly influenced by introduction order gen availability and fungivore presence can be illustrated by (Fig. 3u–x) and T. versicolor function was strongly influenced considering the prevalence of B. citrina. Under low-nitrogen by the nitrogen and fungivore treatments (Fig. 1l, r and x), conditions, this species occupied the smallest volume of wood decomposition rate was much higher when T. versicolor when arriving after T. versicolor, but at high nitrogen and in arrived first to colonise a large portion of the wood, but this the presence of fungivores, had its highest occupancy when effect was reduced by nitrogen addition and fungivore pres- arriving after the same species (Fig. 3e–h). The exact mecha- ence. The introduction order of T. versicolor was also largely nisms remain uncertain, but increased nitrogen availability

Treatment (Dickie et al. 1998). Alternatively, priority effects can be the (a) − Nitrogen result of niche modification by early-arriving species. There is − Fungivores some evidence that the strength of priority effects for wood- − Nitrogen decomposing fungi is linked to their ability to degrade lignocel- + Fungivores lulose (Cline & Zak 2015; Hiscox et al. 2015a,b), presumably + Nitrogen as a result of biochemical modification of the substrate mate- − Fungivores rial. In our study, initial colonisation by T. versicolor, a ligni- + Nitrogen nolytic, white-rot fungus, led to a distinct species composition, + Fungivores but this effect was less apparent when nitrogen was added (Fig. 2a). Cline & Zak (2015) observed a similar phenomenon, Initial species reporting that priority effects exerted by ligninolytic fungi were NMDS axis 2 Trametes stronger on low-nutrient, high-lignin leaf litter and weaker on more nutrient-rich litter. However, these observations may Ascocoryne reflect either a change in the relative impact of the initial colo- niser, due to a change in extracellular enzyme production (Lea- Bisporella tham & Kirk 1983), or greater nutrient availability facilitating colonisation by later arriving species. Daldinia For wood decomposition, we found that the strength of priority effects was modified by the interactive effects of nitrogen NMDS axis 1 availability and fungivore presence, primarily due to the (b) response of T. versicolor to the experimental treatments. This species induced a relatively high decomposition rate compared 2 to the other species, but this distinction was diminished by fungivore presence and nearly eliminated by nitrogen addition 1 (Fig. 1l). For some wood-decomposing fungi, nitrogen supplementation can inhibit or delay ligninolytic enzyme production, resulting in lower decomposition rates (Keyser et al. 1978; Lea- 0 Distance to centroid − Nitrogen − Nitrogen + Nitrogen + Nitrogen tham & Kirk 1983). Grazing by fungivores has also been shown − Fungivores + Fungivores − Fungivores + Fungivores to affect extracellular enzyme production and decomposition rate (Crowther et al. 2011c,d). Both nitrogen and fungivores Figure 2 (a) Non-metric multidimensional scaling (NMDS) of fungal may have affected decomposition by modifying enzyme produc- community composition (stress = 0.17), using the modified Gower tion in T. versicolor, but the stronger effect of nitrogen addition dissimilarity index proposed by Anderson et al. (2006). Colours indicate may have precluded any additional effect of fungivory. Alterna- assembly history treatment, fill (empty or solid) indicates the omission or tively, if fungivores reduced decomposition rate by requiring addition of supplemental nitrogen and shape (square vs. circle) indicates the presence or absence of fungivores. (b) Compositional dissimilarity that the fungus reallocate resources to defence or to repair dam- (modified Gower) for each combination of nitrogen and fungivore age caused by grazing, nitrogen supplementation may simply treatments, calculated using the betadisper function in the R-package satisfy the additional demand imposed by fungivore grazing. vegan. Large points indicate the mean distance to the multivariate group The ligninolytic capabilities of T. versicolor may largely centroid and small points (coloured by initial species treatment) indicate explain the differences in decomposition rate among assembly the distance for individual samples. history treatments, but our results suggest that this species was also more adept at translocating available nitrogen than may have altered the effects of fungivores by altering the com- the other species: highest mean nitrogen concentrations at 6 position of the fungal community that the Collembola were months were observed when T. versicolor arrived first interacting with (Kardol et al. 2016; Morrison et al. 2016); by (Fig. S6e–h), but at 12 months these treatments had the low- facilitating defensive responses to fungivore grazing, such as est mean nitrogen concentrations (Fig. 4e–h). These changes the production of nitrogen-rich secondary metabolites in nitrogen concentration are not consistent with mass loss to (Crowther et al. 2011b); or by altering hyphal growth rate respiration and suggest that T. versicolor transported nitrogen and morphology (Dickie et al. 1998). from the soil to the wood substrate during the early stages of Although greater resource availability can strengthen priority decomposition and away during the later stages. However, effects (Chase 2003, 2010; Fukami 2015), we found that nitro- these patterns varied with nitrogen addition and fungivore gen addition weakened priority effects for species composition. presence and were not directly correlated with decomposition In this system, nitrogen addition appears to have reduced the rate. In addition, patterns of nitrogen concentration in the inhibition of late-arriving species by early-arriving ones by mixed-species treatments did not directly reflect the patterns allowing more species to occupy a larger portion of the wood observed in the single-species treatments (Fig. 1m–x). These substrate (Fig. 3). There are several possible explanations for discrepancies suggest that nitrogen dynamics during decompo- this observation. For example, if priority effects are primarily sition were influenced by interactions between initial nitrogen the result of niche preemption, nitrogen addition may have availability, fungivore presence and interspecific fungal inter- modified hyphal morphology by increasing local hyphal density actions, possibly involving the production of nitrogen-rich sec- while reducing the spatial extent of initial resource exploration ondary defence compounds or modified, combative hyphae

− Nitrogen − Nitrogen + Nitrogen + Nitrogen − Fungivores + Fungivores − Fungivores + Fungivores (a) (b) (c) (d) 100 H*** 75 N*** F** 50 H x N*** H x F*** Ascocoryne 25 N x F** 0 H x N x F*** (e) (f) (g) (h) 100

75 H*** 50 N***

Bisporella F*** 25 H x N*** 0 H x F** (i) (j) (k) (l) 100

75 H*** N*** 50 F***

Calocera H x N*** 25 H x F*** 0 N x F* (m) (n) (o) (p) 100 H*** 75 N***

% substrate occupied F 50 H x N***

Daldinia H x F** 25 N x F** 0 H x N x F** (q) (r) (s) (t) 100

Sistotrema H*** 25 N*** 0 H x N** (u) (v) (w) (x) 100

50 Trametes 25

0 H*** Daldinia Daldinia Daldinia Daldinia Trametes Trametes Trametes Trametes Bisporella Bisporella Bisporella Bisporella Ascocoryne Ascocoryne Ascocoryne Ascocoryne Initial species

Figure 3 Effect of assembly history (H), nitrogen addition (N) and fungivore presence (F) on the abundance (mean % sub-samples occupied Æ SE) of individual fungal species, excluding species that were not observed in more than one sample. Results are show for Ascocoryne sarcoides (a-d), Bisporella citrina (e-h), Calocera sinensis (i-l), Daldinia novae-zelandiae (m-p), Sistotrema brinkmannii (q-t), and Trametes versicolor (u-x). Predictors remaining after model selection and their signiﬁcance levels are shown for individual binomial generalised linear models, where *P < 0.05, **P < 0.01, ***P < 0.001.

− Nitrogen − Nitrogen + Nitrogen + Nitrogen − Fungivores + Fungivores − Fungivores + Fungivores (a) (b) (c) (d)

30 H *** N *** 20 F *** H x N *** H x F ***

loss (%) 10

Wood mass N x F * 0 H x N x F *** (e) (f) (g) (h) 0.20 H *** N *** 0.15 F 0.10 H x N *** H x F * 0.05 N x F * Nitrogen (%) 0.00 H x N x F ** (i) (j) (k) (l) 3000 H *** N *** 2000 F H x N *** C:N 1000 H x F * N x F * 0 H x N x F ** Daldinia Daldinia Daldinia Daldinia Trametes Trametes Trametes Trametes Bisporella Bisporella Bisporella Bisporella Ascocoryne Ascocoryne Ascocoryne Ascocoryne Initial species

Figure 4 Effect of assembly history (H), nitrogen addition (N) and fungivore presence (F) on wood substrate characteristics 12 months after inoculation. Bar heights for all response variables (dry mass [a–d], % nitrogen [e–h] and carbon-to-nitrogen ratio [i–l]) indicate mean Æ SE. Horizontal dotted lines indicate the mean % nitrogen or carbon-to-nitrogen ratio of the non-inoculated controls. Signiﬁcant predictors are indicated for each response, where *P < 0.05, **P < 0.01, ***P < 0.001.

(Hedlund et al. 1991; Heilmann-Clausen & Boddy 2005; experimental ease, but inoculation with pre-colonised wood Crowther et al. 2012). blocks (e.g. Boddy 2000) may have yielded different results Our study used a small number of fungal species and assem- and may better represent scenarios in which wood is colonised bly history treatments so that the strength of priority effects by mycelia from neighbouring wood substrates. However, was primarily attributed to the influence of the experimental recent field studies suggest that the types of priority effects treatments on one species, T. versicolor. A larger species pool observed in our 1-year experiment may have long-term conse- may include a broader spectrum of functionally diverse species quences for fungal and arthropod community assembly and and may also increase the potential for priority effects by wood decomposition (Weslien et al. 2011; Ottosson et al. increasing the probability that some species will have high 2014). Long-term priority effects may be important to consider niche overlap (Fukami 2015). More data on fungal traits than for understanding how climate and land-use change affect currently available would facilitate testing of this hypothesis. wood decomposition as climate variability (Kauserud et al. Similarly, we used only one fungivore species to introduce 2012; Diez et al. 2013; Boddy et al. 2014) and the size and top-down pressure, but fungi are likely to be consumed by a connectivity of forest reserves (Abrego et al. 2015) alter fungal wide range of organisms in natural systems. Recent evidence phenology. Furthermore, knowledge of long-term priority suggests that fungivorous macrofauna in soils, such as isopods effects and their joint regulation by top-down and bottom-up or millipedes, can have greater effects on fungal community forces may help to design effective reintroduction of threat- assembly than Collembola and other mesofauna (Crowther ened fungal species for conservation (Abrego et al. 2016). et al. 2011a, 2013). In this sense, our use of Collembola might Although we have focused here on a simple community of be viewed as a conservative test of top-down effects. However, wood-decomposing fungi, interactive regulation of priority depending on the level of specialisation and functional differ- effects by top-down and bottom-up forces may be common in ences among fungivores, multiple species of fungivores may natural systems. For example, Bakker et al. (2006) demon- cancel out one another’s effect on community assembly. strated that the effects of herbivores on plant diversity in Future efforts to extend our results could incorporate mesh grasslands can vary from negative to positive with increasing bags to exclude different size-classes of soil fauna in field- resource availability, suggesting that the joint effects of top- based experimental manipulations (Dickie et al. 2012). down and bottom-up factors may affect priority effects in It remains uncertain how applicable the results from our these communities. In addition, if herbivores preferentially experiment are to natural community assembly, given the arti- consume more productive species, or certain plant functional ficial conditions. For example, we used agar plugs for groups, they could modify the effects of assembly history and

© 2017 John Wiley & Sons Ltd/CNRS Letter Grazing, resources, and priority effects 9 local resource availability on ecosystem function (Ejrnæs et al. Boberg, J.B., Finlay, R.D., Stenlid, J., Ekblad, A. & Lindahl, B.D. 2006). Similarly, in aquatic invertebrate communities the (2014). Nitrogen and carbon reallocation in fungal mycelia during strength of priority effects on community composition is also decomposition of boreal forest litter. PLoS ONE, 9, e92897. Boddy, L. (2000). Interspecific combative interactions between wood- known to vary with resources (Chase 2010) and predation decaying basidiomycetes. FEMS Microbiol. Ecol., 31, 185–194. (Louette & De Meester 2007; Chase et al. 2009). If resource Boddy, L., Buntgen,€ U., Egli, S., Gange, A.C., Heegaard, E., Kirk, P.M. availability also influences the abundance of predators in these et al. (2014). Climate variation effects on fungal fruiting. Fungal Ecol., systems, interactive control of the strength of priority effects 10, 20–33. may be likely. Interactive effects could also arise if predator Boddy, L., Hiscox, J., Gilmartin, E.C., Johnston, S.R. & Heilmann- presence limits prey species’ access to resources, indirectly Clausen, J. (2017). Wood decay communities in angiosperm wood. In: modifying resource availability. We suggest that further The Fungal Community: Its Organization and Role in the Ecosystem (eds Dighton, J. & White, J.F.). Fourth Edition. New York, NY, CRC efforts to understand how multiple factors interact to regulate Press, pp. 169–190. priority effects will contribute to identifying the conditions Booth, R.G. & Anderson, J.M. (1979). The influence of fungal food that make assembly history important to community structure quality on the growth and fecundity of Folsomia candida (Collembola: and function. Isotomidae). Oecologia, 38, 317–323. Chase, J.M. (2003). Community assembly: when should history matter? Oecologia, 136, 489–498. ACKNOWLEDGEMENTS Chase, J.M. (2010). Stochastic community assembly causes higher – We thank Duckchul Park, Karyn Hoksbergen, and Barbara biodiversity in more productive environments. Science, 328, 1388 1391. Chase, J.M., Biro, E.G., Ryberg, W.A. & Smith, K.G. (2009). Predators Paulus for laboratory assistance, Peter Johnston for advice, temper the relative importance of stochastic processes in the assembly Shenandoah Forest for donation of wood, and three anony- of prey metacommunities. Ecol. Lett., 12, 1210–1218. mous reviewers for comments. This work was supported by Cline, L.C. & Zak, D.R. (2015). Initial colonization, community assembly the Marsden Fund using New Zealand Government funding and ecosystem function: fungal colonist traits and litter biochemistry administered by the Royal Society of New Zealand (contract mediate decay rate. Mol. Ecol., 24, 5045–5058. LCR0503) and the former New Zealand Foundation for Clinton, P.W., Allen, R.B. & Davis, M.R. (2002). Nitrogen storage and Research, Science and Technology (Ecosystem Resilience availability during stand development in a New Zealand Nothofagus forest. Can. J. For. Res., 32, 344–352. Outcome-Based Investment; contract C09X0502). DRL was Cragg, R.G. & Bardgett, R.D. (2001). How changes in soil faunal supported by the Mycological Society of America Graduate diversity and composition within a trophic group influence Research Fellowship. decomposition processes. Soil Biol. Biochem., 33, 2073–2081. Crowther, T.W., Boddy, L. & Jones, T.H. (2011a). Outcomes of fungal interactions are determined by soil invertebrate grazers. Ecol. Lett., 14, AUTHORSHIP 1134–1142. TF designed the study with input from RBA, IAD, JPW, and Crowther, T.W., Boddy, L. & Jones, T.H. (2011b). Species-specific effects of soil fauna on fungal foraging and decomposition. Oecologia, 167, PKB. JPW led implementation of the experiment. IAD devel- 535–545. oped molecular identification. DRL performed data analysis Crowther, T.W., Jones, T.H. & Boddy, L. (2011c). Species-specific effects with input from IAD and TF and wrote the manuscript with of grazing invertebrates on mycelial emergence and growth from woody input from all other authors. resources into soil. Fungal Ecol., 4, 333–341. Crowther, T.W., Jones, T.H., Boddy, L. & Baldrian, P. (2011d). Invertebrate grazing determines enzyme production by basidiomycete DATA ACCESSIBILITY STATEMENT fungi. Soil Biol. Biochem., 43, 2060–2068. Crowther, T.W., Boddy, L. & Hefin Jones, T. (2012). Functional and Data available from the Dryad Digital Repository: http://dx. ecological consequences of saprotrophic fungus-grazer interactions. doi.org/10.5061/dryad.7p2cv ISME J., 6, 1992–2001. Crowther, T., Stanton, D. & Thomas, S. (2013). Top-down control of soil fungal community composition by a globally distributed keystone REFERENCES consumer. Ecology, 94, 2518–2528. Abrego, N., Bassler,€ C., Christensen, M. & Heilmann-Clausen, J. (2015). Davis, M.R., Allen, R.B. & Clinton, P.W. (2004). The influence of N Implications of reserve size and forest connectivity for the conservation addition on nutrient content, leaf carbon isotope ratio, and of wood-inhabiting fungi in Europe. Biol. Conserv., 191, 469–477. productivity in a Nothofagus forest during stand development. Can. J. – Abrego, N., Oivanen, P., Viner, I., Norden, J., Penttila,€ R., Dahlberg, A. For. Res., 34, 2037 2048. et al. (2016). Reintroduction of threatened fungal species via Diamond, J.M. (1975). Assembly of species communities. In Ecology and inoculation. Biol. Conserv., 203, 120–124. Evolution in Communities (eds Cody, M.L., Diamond, J.M.). Harvard – Allen, R.B., Buchanan, P.K., Clinton, P.W. & Cone, A.J. (2000). University Press, Cambridge, MA, pp. 342 444. Composition and diversity of fungi on decaying logs in a New Zealand Dickie, I.A., Koide, R.T. & Stevens, C.M. (1998). Tissue density and temperate beech (Nothofagus) forest. Can. J. For. Res., 30, 1025–1033. growth response of ectomycorrhizal fungi to nitrogen source and – Allison, S.D., LeBauer, D.S., Ofrecio, M.R., Reyes, R., Ta, A.-M. & concentration. Mycorrhiza, 8, 145 148. Tran, T.M. (2009). Low levels of nitrogen addition stimulate Dickie, I.A., Richardson, S.J. & Wiser, S.K. (2009). Ectomycorrhizal decomposition by boreal forest fungi. Soil Biol. Biochem., 41, 293–302. fungal communities and soil chemistry in harvested and unharvested – Anderson, M.J., Ellingsen, K.E. & McArdle, B.H. (2006). Multivariate temperate Nothofagus rainforests. Can. J. For. Res., 39, 1069 1079. dispersion as a measure of beta diversity. Ecol. Lett., 9, 683–693. Dickie, I.A., Fukami, T., Wilkie, J.P., Allen, R.B. & Buchanan, P.K. Bakker, E.S., Ritchie, M.E., Olff, H., Milchunas, D.G. & Knops, J.M.H. (2012). Do assembly history effects attenuate from species to ecosystem – (2006). Herbivore impact on grassland plant diversity depends on properties? A field test with wood-inhabiting fungi. Ecol. Lett., 15, 133 habitat productivity and herbivore size. Ecol. Lett., 9, 780–788. 141.

Diez, J.M., James, T.Y., McMunn, M. & Ibanez,~ I. (2013). Predicting Morin, P.J. (1984). Odonate guild composition: experiments with species-specific responses of fungi to climatic variation using historical colonization history and fish predation. Ecology, 65, 1866–1873. records. Glob. Change Biol., 19, 3145–3154. Morrison, E.W., Frey, S.D., Sadowsky, J.J., van Diepen, L.T.A., Drake, J. (1991). Community-assembly mechanics and the structure of an Thomas, W.K. & Pringle, A. (2016). Chronic nitrogen additions experimental species ensemble. Am. Nat., 137, 1–26. fundamentally restructure the soil fungal community in a temperate Ejrnæs, R., Bruun, H.H. & Graae, B.J. (2006). Community assembly in forest. Fungal Ecol., 23, 48–57. experimental grasslands: suitable or timely arrival? Ecology, 87, 1225– Oksanen, J., Blanchet, F.G., Friendly, M., Kindt, R., Legendre, P., 1233. McGlinn, D. et al. (2017). vegan: Community Ecology Package. R FitzJohn, R.G. & Dickie, I.A. (2007). TRAMPR: an R package for package version 2.4-2. Available at: https://CRAN.R-project.org/packa analysis and matching of terminal-restriction fragment length ge=vegan. polymorphism (TRFLP) profiles. Mol. Ecol. Notes, 7, 583–587. Ottosson, E., Norden, J., Dahlberg, A., Edman, M., Jonsson,€ M., Fountain, M.T. & Hopkin, S.P. (2005). Folsomia candida (Collembola): a Larsson, K.-H., et al. (2014). Species associations during the succession “standard” soil arthropod. Annu. Rev. Entomol., 50, 201–222. of wood-inhabiting fungal communities. Fungal Ecol., 11, 17–28. Fukami, T. (2015). Historical contingency in community assembly: Orrock, J.L. & Fletcher, R.J. Jr (2005). Changes in community size affect integrating niches, species pools, and priority effects. Annu. Rev. Ecol. the outcome of competition. Am. Nat., 166, 107–111. Evol. Syst., 46, 1–23. Palmgren, A. (1926). Chance as an element in plant geography. In: Fukami, T., Dickie, I.A., Wilkie, J.P., Paulus, B.C., Park, D., Roberts, A. Proceedings of the International Congress of Plant Sciences (ed. Duggar, et al. (2010). Assembly history dictates ecosystem functioning: evidence B.). Ithica, New York, George Banta Publishing Company, pp. 591– from wood decomposer communities. Ecol. Lett., 13, 675–684. 602. Fukami, T., Mordecai, E.A. & Ostling, A. (2016). A framework for Pinheiro, J., Bates, D., DebRoy, S. & Sarkar, D. & R Core Team. (2016). priority effects. J. Veg. Sci., 27, 655–657. nlme: Linear and Nonlinear Mixed Effects Models. R package version Hedlund, K., Boddy, L. & Preston, C.M. (1991). Mycelial responses of 3.1-130. Available at: https://CRAN.R-project.org/package=nlme. the soil fungus, Mortierella isabellina, to grazing by Onychiurus armatus R Development Core Team. (2016). R: A Language and Environment for (Collembola). Soil Biol. Biochem., 23, 361–366. Statistical Computing. R Foundation for Statistical Computing, Heilmann-Clausen, J. & Boddy, L. (2005). Inhibition and stimulation Vienna, Austria. effects in communities of wood decay fungi: exudates from colonized Smaill, S.J., Clinton, P.W., Allen, R.B. & Davis, M.R. (2011). Climate wood influence growth by other species. Microb. Ecol., 49, 399–406. cues and resources interact to determine seed production by a masting Hiscox, J., Savoury, M., Muller,€ C.T., Lindahl, B.D., Rogers, H.J. & species. J. Ecol., 99, 870–877. Boddy, L. (2015a). Priority effects during fungal community Sutherland, J.P. (1974). Multiple stable points in natural communities. establishment in beech wood. ISME J.,9,1–15. Am. Nat., 108, 859–873. Hiscox, J., Savoury, M., Vaughan, I.P., Muller,€ C.T. & Boddy, L. Tan, J., Pu, Z., Ryberg, W.A. & Jiang, L. (2012). Species phylogenetic (2015b). Antagonistic fungal interactions influence carbon dioxide relatedness, priority effects, and ecosystem functioning. Ecology, 93, evolution from decomposing wood. Fungal Ecol., 14, 24–32. 1164–1172. Jones, H.L. & Worrall, J.J. (1995). Fungal biomass in decayed wood. Tucker, C.M. & Fukami, T. (2014). Environmental variability counteracts Mycologia, 87, 459–466. priority effects to facilitate species coexistence: evidence from nectar Kardol, P., Souza, L. & Classen, A.T. (2013). Resource availability microbes. Proc. R. Soc. B Biol. Sci., 281, 20132637. mediates the importance of priority effects in plant community Vannette, R.L. & Fukami, T. (2014). Historical contingency in species assembly and ecosystem function. Oikos, 122, 84–94. interactions: towards niche-based predictions. Ecol. Lett., 17, 115–124. Kardol, P., Spitzer, C.M., Gundale, M.J., Nilsson, M.-C. & Wardle, D.A. Wardle, D.A. (2002). Communities and Ecosystems: Linking the (2016). Trophic cascades in the bryosphere: the impact of global change Aboveground and Belowground Components. Princeton University Press, factors on top-down control of cyanobacterial N2-fixation. Ecol. Lett., Princeton, New Jersey. 19, 967–976. Wardle, D.A. & Yeates, G.W. (1993). The dual importance of Kauserud, H., Heegaard, E., Buntgen,€ U., Halvorsen, R., Egli, S., Senn- competition and predation as regulatory forces in terrestrial ecosystems: Irlet, B. et al. (2012). Warming-induced shift in European mushroom evidence from decomposer food-webs. Oecologia, 93, 303–306. fruiting phenology. PNAS, 109, 14488–14493. Weslien, J., Djupstrom,€ L.B., Schroeder, M. & Widenfalk, O. (2011). Keyser, P., Kirk, T.K. & Zeikus, J.G. (1978). Ligninolytic enzyme system Long-term priority effects among insects and fungi colonizing decaying of Phanerochaete chrysosporium: synthesized in the absence of lignin in wood. J. Anim. Ecol., 80, 1155–1162. response to nitrogen starvation. J. Bacteriol., 135, 790–797. Worrall, J.J., Anagnost, S.E. & Zabel, R.A. (1997). Comparison of wood Korner,€ C., Stocklin,€ J., Reuther-Thiebaud, L. & Pelaez-Riedl, S. (2008). decay among diverse lignicolous fungi. Mycologia, 89, 199–219. Small differences in arrival time influence composition and productivity Zimmerman, J.K., Pulliam, W.M., Lodge, D.J., Quinonesorfila, V., of plant communities. New Phytol., 177, 698–705. Fetcher, N., Guzmangrajales, S. et al. (1995). Nitrogen immobilization Larsen, T., Ventura, M., O’Brien, D.M., Magid, J., Lomstein, B.A. & by decomposing woody debris and the recovery of tropical wet forest Larsen, J. (2011). Contrasting effects of nitrogen limitation and amino from hurricane damage. Oikos, 72, 314–322. acid imbalance on carbon and nitrogen turnover in three species of Collembola. Soil Biol. Biochem., 43, 749–759. Lawton, J.H. (1999). Are there general laws in ecology? Oikos, 84, 177–192. SUPPORTING INFORMATION Leatham, G.F. & Kirk, T.K. (1983). Regulation of ligninolytic activity by Additional Supporting Information may be found online in nutrient nitrogen in white-rot basidiomycetes. FEMS Microbiol. Lett., 16, 65–67. the supporting information tab for this article. Louette, G. & De Meester, L. (2007). Predation and priority effects in experimental zooplankton communities. Oikos, 116, 419–426. Lussenhop, J. (1996). Collembola as mediators of microbial symbiont Editor, John Klironomos – effects upon soybean. Soil Biol. Biochem., 28, 363 369. Manuscript received 20 March 2017 Moore, J.C., McCann, K., Setal€ a,€ H. & De Ruiter, P.C. (2003). Top- down is bottom-up: does predation in the rhizosphere regulate First decision made 26 April 2017 aboveground dynamics? Ecology, 84, 846–857. Manuscript accepted 4 June 2017